The CLARA climate data record is the longest remote sensing based record for global surface variables, and has unique significance to environmental and climate studies. This manuscript presents an overview to the latest version of CLARA-A3 albedo product, including its algorithm update and features of dataset. The content is relevant to the scope of ESSD, and will be helpful to data users. The aim of this manuscript is to clearly describe the strength and limitations of the dataset. In this consideration, some questions and revision suggestions are raised as follows:

1. Firstly, I would complain that data is not directly accessable at the provided DOI: 5676/EUM_SAF_CM/CLARA_AVHRR/V003.
2. In table 1, I just ask a question: is there any change in the algorithm of albedo retrieval over ocean surface?
3. In Line 98, page 5, “As AVHRR geolocation is calculated on a geodesic reference ellipsoid, a combination of sufficiently large elevation and viewing angle requires across-track shifting of pixels to obtain true geolocation”, I cannot easily understand the phase “requires across-track shifting of pixels to obtain true geolocation”. And I think maybe the geometric processing of AVHRR data is not in the scope of this manuscript, otherwise, it would be too much to discuss in a single article.
4. In Line 100, page 5, it is not clear whether the mean slope refers to slope of 30m pixel or 0.05DEG pixel. And whether the mean refers to the mean of 0.05DEG pixel or 0.25DEG pixel. If possible, please indicate the percentage of pixels which need topology correction.
5. In Line 103, page 5, it is not clear whether the “each AVHRR pixel to be corrected” refers to 05DEG pixel or 0.25DEG pixel.
6. In line 123, page 6, it is not clear whether the same narrow-to-broadband conversion (NTBC) algorithm in Liang et al. (2000) is applied both for snow-ice surface and snow-free surface.
7. In line 231, page 9, “available sampling may change by a factor of 50”, what is the meaning of “by a factor of 50”?
8. In line 236, page 9, the “climatological albedo” is used in this manuscript as a reference data. But the source and accuracy of climatological albedo cannot be found throughout the manuscript.
9. Figure 4b, pages 10. I found it difficult to understand this subfigure. Are the scattered points in Figure 4b a single time sequence? If so, why are they not continuous as in Figure 4a?
10. In line 284, page 12, “the decadal stability of bias, i.e. the temporal trend in bias as per cent per decade”. The use of temporal trend as stability indicator is not questionable to me. It is possible the long-term trend is very small while the sequence of albedo looks very noisy and unstable. I would suggest using variance of the bias as stability indicator.
11. Figure 6c, page 16, “Sizes of rectangular markers indicate the amount of valid clear-sky AVHRR data of each month.” But I cannot find the rectangular markers in the figure.

This manuscript discribed a long time series CLARA albedo climate data record from 1979 to present. This dataset is esential to environmental supervision and climate change studies. This study presents the overview of the retrieval algorithm and the validation of the new released CLARA albedo products. The topic is interesting to the the journal of ESSD. However, several issues existed and made the manuscript hard to understand. My major concerns are following:

• What’s the main improvement of the new released albedo products? The high quality, the higher spatial resolution or the more frequency?
• Up to now, I could not download your datasets (using the link as you provided DOI). How can you guarantee data access on the NTP data center servers?
• I suggest separating the contents of “data record description” from the current of “Data record description and algorithm overview”.
• There are too much describtion about the albedo retrieval algorithm. I suggest the author make it more logical.
• Does the retrieval algorithm have some difference with the proposed aogorithm?
• Does the NTB method can be directly used to transfer the narrowband reflectance to broadband albedo? As we know, the NTB method were built following the assumption of the land surface is Lambertian.
• Which method was used to reduce the topographic effects on the imageries? Does the AVHRR imageries have the serious topographic effects at so coarse-scale of 0.25° to 25km?
• Which BRDF correction method was used in this manuscript?
• Does any gap exist in the current new released CLARA albedo products follows your strict selection of observations (such as the Sun Zenith Angle >70 deg. or Viewing Zenith Angle >60 deg)?
• How to screen out the snow-covered pixels for the retrieval of albedo over the global scale.
• I can’t understand the sentence “As AVHRR geolocation is calculated on a geodesic reference ellipsoid, a combination of sufficiently large elevation and viewing angle requires across-track shifting of pixels to obtain true geolocation” in Line 98-99 in page 5. Could you please explain it and make it more eaily to understand for the readers?
• Does the SRTM have the DEM data between the surface that lies between the 56 degrees south to 60 degrees south latitude? As we all know, the SRTM’s radars can cover most of Earth's land surface that lies between 60 degrees north and 56 degrees south latitude.
• What is the meaning of the word of “TOMS” in line L111, page 5? I can’t find any describtion about this word.
• I find that you used the AOD data from 2015 to 2014 to replace the AOD data from 2005 to 2014. How many uncecessary uncertainties were introduced?
• There are many Land Cover products, such as the USGS land cover products, the GLC2000 product, and so on. Do they have the same land cover classification principles? Do they have the same spatial resolution? If they have the unsame land cover classifications, how to merge them?
• Are the same albedo retrieval algorithms in the open wate and the closed water?
• I can’t understand the sentence “Note that the AVHRR-observed reflectances are not used in the estimation of ocean surface albedo.” Why the AVHRR reflectance can not be used for generating land surface albedo?
• Does the statistics parameter used in Eq (2) and (3) can be used in the global scale?
• How to calculate the diffuse and direct radiation? How to calculate the CP? Which data can be used?
• I suggest to more quantitative assessment of the new released CLARA albedo. The scatter plots can be used here both in the snow-covered surface and snow-free land surface.
• Which method was used to assess the representastiveness of the site?
• I found that many sentences abou the quality of the new released albedos do not have the necessary figures or the tables to support, such as the sentence from 261 to 264, the sentence from 256 to 260, Line 265-274.
• How to select the cloud-free in situ albedo measurements to validation of the new released albedo product?
• I suggest the author giving a scatter plot between the CLARA albedo and MCD43D51 in the global scale to show the accuracy of the CLARA albedo.
• I found that the larger biases exist between the CLARA albedo and MCD43D51 albedo, especially over theSouthern Hemisphere. What are reasons?
• The DPI fo the figures need to be improved.

The paper presents the third release of the CLARA surface albedo climate data record. The CDR covers four decades.  Such dataset is definitely of great value, in particular over the arctic region.

Due to lack of measurements is less robust over the tropical belt and this should be highlighted a bit more from the authors. 

The paper is well written, the method and the results clearly expressed even if some parts could be extended with deeper explanation and adding some figures. For instance the section on uncertainty. Also the part on validation should contain more results and analysis. As explained later, I find a bit weak basing the analysis on big spatial and temporal mean values comparison.  

Some comments follows. If the authors keep them into account, the paper for me is suitable for publication. 

Minor issues:

Method: the retrieval method is well known to be robust to generate good retrievals but it does not account for coupling between the surface and the scattering atmosphere close to it. Could the authors expand a bit on this point?

Explain the meaning of the acronyms SAL, WAL, BAL (pag 3 line 63). The three albedos are explained with equations in the relative paragraph, but it is missing a sentence explaining them. Those variables are just not only mathematical terms but have a meaning and I think this must be added, in particular for non-expert readers or data users.

Table 1: add reference for ERA-Interim and ERA5.

Pag. 4 Add a similar plot (Figure 1 a) for a pentad product. This is necessary to give a complete example of the different products provided to the users. This would show if specific difference can be due to the amount of days accumulated.

Pag. 12, Table 2 add a column to define the surface type of each site. It would be of great help for the readers to have a plot showing the location on Earth of the sites.

Pag. 24, Figure 11. This figure has potential for me but the re no clear meaning on what is the message in it. Could the authors spend a couple of sentences to add a deeper explanations?

Major issues:

Pag 3 line 62,63: You introduce the white, black, blue sky albedos. You miss to explain their meaning and their relation. This should be add. See for instance (https://doi.org/10.1175/JAS3479.1)

Pag. 10 Figure 4. I suggest to add a figure similar to (b) but for one region over the tropics. This will show how different geographical regions are impacted in the retrieval and provide the users with a clear view of the potentialities of the products in different areas.   

Pag. 9 Line 219. The authors say that being the retrieval deterministic rather than probabilistic but the measure of uncertainty is left to statistical values. The albedo variables are estimated using clear mathematical relations. Could the authors explain what the uncertainty are not also mathematically estimated using the well known procedures. See for instance the method used in https://doi.org/10.1029/2006JD007313. I find this section personally to be the only one a bit weak in the paper, otherwise well structured.

Section 3 on validation. The authors have a very long list of validation sites. Even if they clearly worked a lot and well on this part, I have several issues with this paragraph. The authors includes a lot of sites (in red italics) not used in the analysis being “spatially” unrepresentative. Could you explain why they have been included if not sued? For the remaining sites, I would have expected some full time range (4 decades or the time covered by the ground measurements at least) plots (selecting some representative among all, covering the different surface types). Only basing the analysis on monthly mean could be misleading for me.  There are some summary statistics in Table 4. The table shows interesting figures. I would recommend to provide the numbers for each valid site and not as merged statistic. 

Section 4 on intercomparison to MODIS. This is the only comparison against satellite data. It is a pity not to intercompare against a geostationary product but I can see the difficulty even if it would have made the section more complete. The comparison against MODIS is a bit weak. There is only one figure, again showing means over months. It would be really important to select some ground sites and show time series including also MODIS and CLARA-3.